1. Field of the Invention
The invention relates to data tracking methods and systems applied in semiconductor manufacturing, and more particularly to split lot tracking methods and systems using data warehouses applied in semiconductor manufacturing.
2. Description of the Related Art
Wafer manufacturing in semiconductor comprises split lot processing for generating new wafer lots, merging lot processing for parent, child, or similar lots, reworking lot processing for reworking procedures, and similar.
With respect to split lot processing, split lot tracking is essential and significant for production management and process analysis in semiconductor fabrications. Currently, a historical lot table is created, in which lot numbers (Lot_ID) and process stations (Ope_No) serve as a primary key. In each wafer data, the lot number records sub lot numbers when a split operation is executed. Database designs of some process management systems utilize denormalization with lot numbers of the previous process station to implement tracking processing. Thousands of data, however, are generated at each process station such that a single data cannot record all the split lot numbers.
As shown in Attachment 1, a historical data lot table 100 displays lot numbers and process stations of currently processed wafer lots, defining a lot key number (Lot_Key_No) and a previous lot key number (PRE_Lot_Key_No). The previous lot key number is a backward pointer, pointing to the lot key number. When a sub lot query is required, a previous historical data lot table is looked up according to the lot key number pointed to by the previous lot key number.
Further, Attachment 2 shows a master lot table 210 and a slave lot table 220, displaying lot numbers and process stations of the currently processed wafer lots. A lot ID (Lot_ID) and a terminal lot ID (TERMINALLot_ID) are defined in slave lot table 220. The lot ID points to the terminal lot ID to represent sub lots, corresponding to the terminal lot ID, of current lots, corresponding to the lot ID. Similarly, when a sub lot query is required, a previous historical data lot table is looked up according to the lot ID and the terminal lot ID.
As described, complete split lot record queries in a database are implemented, lot numbers of the previous process station are recursively read until the original wafer data is located, such that nearly one thousand queries are required to obtain a split lot record. This method is time-consuming, complex, and slow, thus, in-line controls and analysis efficiency suffer.
Thus, a split lot tracking method and system using data warehouses capable of rapid and detailed queries of required production data applied in semiconductor manufacturing is desirable.